project3

In your final repo, there should be an R markdown file that organizes all computational steps for evaluating your proposed Facial Expression Recognition framework.

This file is currently a template for running evaluation experiments. You should update it according to your codes but following precisely the same structure.

if(!require("EBImage")){
  install.packages("BiocManager")
  BiocManager::install("EBImage")
}

程辑包‘EBImage’是用R版本4.0.3 来建造的

if(!require("R.matlab")){
  install.packages("R.matlab")
}
if(!require("readxl")){
  install.packages("readxl")
}

if(!require("dplyr")){
  install.packages("dplyr")
}
if(!require("readxl")){
  install.packages("readxl")
}

if(!require("ggplot2")){
  install.packages("ggplot2")
}

if(!require("caret")){
  install.packages("caret")
}

if(!require("glmnet")){
  install.packages("glmnet")
}

if(!require("WeightedROC")){
  install.packages("WeightedROC")
}

if(!require("randomForest")){
  install.packages("randomForest")
}
if(!require("pROC")){
  install.packages("pROC")
}
library(R.matlab)
library(readxl)
library(dplyr)
library(EBImage)
library(ggplot2)
library(caret)
library(glmnet)
library(WeightedROC)
library(randomForest)
library(pROC)

Step 0 set work directories

set.seed(2020)
# setwd("~/Project3-FacialEmotionRecognition/doc")
# here replace it with your own path or manually set it in RStudio to where this rmd file is located. 
# use relative path for reproducibility

Provide directories for training images. Training images and Training fiducial points will be in different subfolders.

train_dir <- "../data/train_set/" # This will be modified for different data sets.
train_image_dir <- paste(train_dir, "images/", sep="")
train_pt_dir <- paste(train_dir,  "points/", sep="")
train_label_path <- paste(train_dir, "label.csv", sep="") 

Step 1: set up controls for evaluation experiments.

In this chunk, we have a set of controls for the evaluation experiments.

• (T/F) cross-validation on the training set
• (T/F) reweighting the samples for training set
• (number) K, the number of CV folds
• (T/F) process features for training set
• (T/F) run evaluation on an independent test set
• (T/F) process features for test set

run.cv <- TRUE # run cross-validation on the training set
sample.reweight <- TRUE # run sample reweighting in model training
K <- 5  # number of CV folds
run.feature.train <- TRUE # process features for training set
run.test <- TRUE # run evaluation on an independent test set
run.feature.test <- TRUE # process features for test set

run.cv.randomForest <- TRUE # run cross-validation on the training set for random forest model 
run.train.randomForest <- TRUE # run evaluation on entire train set
run.test.randomForest <- TRUE # run evaluation on an independent test set

Using cross-validation or independent test set evaluation, we compare the performance of models with different specifications. In this Starter Code, we tune parameter lambda (the amount of shrinkage) for logistic regression with LASSO penalty.

lmbd = c(1e-3, 5e-3, 1e-2, 5e-2, 1e-1)
model_labels = paste("LASSO Penalty with lambda =", lmbd)

Step 2: import data and train-test split

#train-test split
info <- read.csv(train_label_path)
n <- nrow(info)
n_train <- round(n*(4/5), 0)
train_idx <- sample(info$Index, n_train, replace = F)
test_idx <- setdiff(info$Index, train_idx)

If you choose to extract features from images, such as using Gabor filter, R memory will exhaust all images are read together. The solution is to repeat reading a smaller batch(e.g 100) and process them.

n_files <- length(list.files(train_image_dir))
image_list <- list()
for(i in 1:100){
   image_list[[i]] <- readImage(paste0(train_image_dir, sprintf("%04d", i), ".jpg"))
}

Fiducial points are stored in matlab format. In this step, we read them and store them in a list.

#function to read fiducial points
#input: index
#output: matrix of fiducial points corresponding to the index
readMat.matrix <- function(index){
     return(round(readMat(paste0(train_pt_dir, sprintf("%04d", index), ".mat"))[[1]],0))
}

#load fiducial points
fiducial_pt_list <- lapply(1:n_files, readMat.matrix)
save(fiducial_pt_list, file="../output/fiducial_pt_list.RData")

Step 3: construct features and responses

• The follow plots show how pairwise distance between fiducial points can work as feature for facial emotion recognition.

  • In the first column, 78 fiducials points of each emotion are marked in order.
  • In the second column distributions of vertical distance between right pupil(1) and right brow peak(21) are shown in histograms. For example, the distance of an angry face tends to be shorter than that of a surprised face.
  • The third column is the distributions of vertical distances between right mouth corner(50) and the midpoint of the upper lip(52). For example, the distance of an happy face tends to be shorter than that of a sad face.

Figure1

feature.R should be the wrapper for all your feature engineering functions and options. The function feature( ) should have options that correspond to different scenarios for your project and produces an R object that contains features and responses that are required by all the models you are going to evaluate later.

• feature.R
• Input: list of images or fiducial point
• Output: an RData file that contains extracted features and corresponding responses

source("../lib/feature.R")
tm_feature_train <- NA
if(run.feature.train){
  tm_feature_train <- system.time(dat_train <- feature(fiducial_pt_list, train_idx))
  save(dat_train, file="../output/feature_train.RData")
}else{
  load(file="../output/feature_train.RData")
}
tm_feature_test <- NA
if(run.feature.test){
  tm_feature_test <- system.time(dat_test <- feature(fiducial_pt_list, test_idx))
  save(dat_test, file="../output/feature_test.RData")
}else{
  load(file="../output/feature_test.RData")
}

Step 4: Train a classification model with training features and responses

Call the train model and test model from library.

train.R and test.R should be wrappers for all your model training steps and your classification/prediction steps.

• train.R
  • Input: a data frame containing features and labels and a parameter list.
  • Output:a trained model
• test.R
  • Input: the fitted classification model using training data and processed features from testing images
  • Input: an R object that contains a trained classifier.
  • Output: training model specification
• In this Starter Code, we use logistic regression with LASSO penalty to do classification.

source("../lib/train.R") 
source("../lib/test.R")

Model selection with cross-validation

• Do model selection by choosing among different values of training model parameters.

source("../lib/cross_validation.R")
feature_train = as.matrix(dat_train[, -6007])
label_train = as.integer(dat_train$label) 
if(run.cv){
  res_cv <- matrix(0, nrow = length(lmbd), ncol = 4)
  for(i in 1:length(lmbd)){
    cat("lambda = ", lmbd[i], "\n")
    res_cv[i,] <- cv.function(features = feature_train, labels = label_train, K, 
                              l = lmbd[i], reweight = sample.reweight)
  save(res_cv, file="../output/res_cv.RData")
  }
}else{
  load("../output/res_cv.RData")
}

lambda =  0.001 
lambda =  0.005 
lambda =  0.01 
lambda =  0.05 
lambda =  0.1 

Visualize cross-validation results.

  
res_cv <- as.data.frame(res_cv) 
colnames(res_cv) <- c("mean_error", "sd_error", "mean_AUC", "sd_AUC")
res_cv$k = as.factor(lmbd)

if(run.cv){
  p1 <- res_cv %>% 
    ggplot(aes(x = as.factor(lmbd), y = mean_error,
               ymin = mean_error - sd_error, ymax = mean_error + sd_error)) + 
    geom_crossbar() +
    theme(axis.text.x = element_text(angle = 90, hjust = 1))
  
  p2 <- res_cv %>% 
    ggplot(aes(x = as.factor(lmbd), y = mean_AUC,
               ymin = mean_AUC - sd_AUC, ymax = mean_AUC + sd_AUC)) + 
    geom_crossbar() +
    theme(axis.text.x = element_text(angle = 90, hjust = 1))
  
  print(p1)
  print(p2)
}

NA
NA

• Choose the “best” parameter value

par_best <- lmbd[which.min(res_cv$mean_error)] # lmbd[which.max(res_cv$mean_AUC)]

• Train the model with the entire training set using the selected model (model parameter) via cross-validation.

# training weights
weight_train <- rep(NA, length(label_train))
for (v in unique(label_train)){
  weight_train[label_train == v] = 0.5 * length(label_train) / length(label_train[label_train == v])
}
if (sample.reweight){
  tm_train <- system.time(fit_train <- train(feature_train, label_train, w = weight_train, par_best))
} else {
  tm_train <- system.time(fit_train <- train(feature_train, label_train, w = NULL, par_best))
}
save(fit_train, file="../output/fit_train.RData")

Step 5: Run test on test images

tm_test = NA
feature_test <- as.matrix(dat_test[, -6007])
if(run.test){
  load(file="../output/fit_train.RData")
  tm_test <- system.time({label_pred <- as.integer(test(fit_train, feature_test, pred.type = 'class')); 
                          prob_pred <- test(fit_train, feature_test, pred.type = 'response')})
}

• evaluation

## reweight the test data to represent a balanced label distribution
label_test <- as.integer(dat_test$label)
weight_test <- rep(NA, length(label_test))
for (v in unique(label_test)){
  weight_test[label_test == v] = 0.5 * length(label_test) / length(label_test[label_test == v])
}

accu <- sum(weight_test * (label_pred == label_test)) / sum(weight_test)
tpr.fpr <- WeightedROC(prob_pred, label_test, weight_test)
auc <- WeightedAUC(tpr.fpr)


cat("The accuracy of model:", model_labels[which.min(res_cv$mean_error)], "is", accu*100, "%.\n")

The accuracy of model: LASSO Penalty with lambda = 0.005 is 71.5625 %.

cat("The AUC of model:", model_labels[which.min(res_cv$mean_error)], "is", auc, ".\n")

The AUC of model: LASSO Penalty with lambda = 0.005 is 0.7956944 .

Summarize Running Time

Prediction performance matters, so does the running times for constructing features and for training the model, especially when the computation resource is limited.

cat("Time for constructing training features=", tm_feature_train[1], "s \n")

Time for constructing training features= 1.237 s 

cat("Time for constructing testing features=", tm_feature_test[1], "s \n")

Time for constructing testing features= 0.353 s 

cat("Time for training model=", tm_train[1], "s \n") 

Time for training model= 3.853 s 

cat("Time for testing model=", tm_test[1], "s \n")

Time for testing model= 0.102 s 

####Random Forest ### Step 1-3(additional):

ntree = c(50, 100, 150,200,250)
randomForest_model_labels = paste("Random Forest with number of trees =", ntree)

Step 4: Train a classification model with training features and responses

source("../lib/randomForest.R") 

feature_train = as.matrix(dat_train[, -6007])
label_train = as.integer(dat_train$label) 

if(run.cv.randomForest){
  res_cv_randomForest <- matrix(0, nrow = length(ntree), ncol = 2)
  for (i in 1:length(ntree)){
cat("ntree =", ntree[i],"\n")
res_cv_randomForest[i,] <- cv.randomForest.function(features = feature_train, labels = label_train,K,ntree = ntree[i])
}
  save(res_cv_randomForest, file="../output/res_cv_randomForest.RData")
}else{
  load("../output/res_cv_randomForest.RData")
}

ntree = 50 
ntree = 100 
ntree = 150 
ntree = 200 
ntree = 250 

res_cv_randomForest <- as.data.frame(res_cv_randomForest)
colnames(res_cv_randomForest) <- c("mean_error", "sd_error")
res_cv_randomForest$ntree = as.integer(ntree)
res_cv_randomForest

if(run.cv.randomForest){
plot_meanError_randomForest <- res_cv_randomForest %>%
ggplot(aes(x = as.factor(ntree), y = mean_error,
ymin = mean_error - sd_error, ymax = mean_error + sd_error)) +
geom_crossbar() +
theme(axis.text.x = element_text(angle = 90, hjust = 1))
}
print(plot_meanError_randomForest)

if(run.cv.randomForest){
  ntree_best_randomForest <- res_cv_randomForest$ntree[which.min(res_cv_randomForest$mean_error)]
}
save(ntree_best_randomForest,file = "../output/ntree_best_randomForest.Rdata")
cat("ntree_best_randomForest=", ntree_best_randomForest)

ntree_best_randomForest= 50

###Step 5: Run test on test images

##Training
tm_train=NA
tm_train <- system.time(fit_train_randomForest <- train_randomForest(feature_train, label_train, ntree=ntree_best_randomForest))
save(fit_train_randomForest, file="../output/fit_train_randomForest.RData")

##Testing
tm_test=NA
if(run.test.randomForest){
  load(file="../output/fit_train_randomForest.RData")
  tm_test <- system.time(label_pred_randomForest <- ifelse(test_randomForest(fit_train_randomForest, feature_test)== 2, 1, 0))
}

• evaluation

test_label <- dat_test$label
accurancy.randomForest <- mean(test_label == label_pred_randomForest)
tpr.fpr <- WeightedROC(as.numeric(label_pred_randomForest), test_label)
auc.randomForest <- WeightedAUC(tpr.fpr)

cat("The accuracy of random forest model:", randomForest_model_labels[which.min(res_cv_randomForest$mean_error)], "is", accurancy.randomForest*100, "%.\n")

The accuracy of random forest model: Random Forest with number of trees = 50 is 81.16667 %.

cat("The AUC of random forest model:", randomForest_model_labels[which.min(res_cv_randomForest$mean_error)], "is", auc.randomForest, ".\n")

The AUC of random forest model: Random Forest with number of trees = 50 is 0.5416667 .

Summarize Running Time

Prediction performance matters, so does the running times for constructing features and for training the model, especially when the computation resource is limited.

cat("Time for constructing training features=", tm_feature_train[1], "s \n")

Time for constructing training features= 1.237 s 

cat("Time for constructing testing features=", tm_feature_test[1], "s \n")

Time for constructing testing features= 0.353 s 

cat("Time for training random forest model=", tm_train[1], "s \n") 

Time for training random forest model= 66.237 s 

cat("Time for testing random forest model=", tm_test[1], "s \n")

Time for testing random forest model= 0.136 s 

###Reference - Du, S., Tao, Y., & Martinez, A. M. (2014). Compound facial expressions of emotion. Proceedings of the National Academy of Sciences, 111(15), E1454-E1462.

####Random Forest with weight ### Step 1-3(additional):

run.cv.randomForestWeight <- TRUE # run cross-validation on the training set for random forest with weight model 
run.train.randomForestWeight <- TRUE # run evaluation on entire train set
run.test.randomForestWeight <- TRUE # run evaluation on an independent test set
ntree = c(50, 100, 150,200,250)
randomForestWeight_model_labels = paste("randomForestWithWeight with number of trees =", ntree)

Step 4: Train a classification model with training features and responses

source("../lib/randomForest.R") 

feature_train = as.matrix(dat_train[, -6007])
label_train = as.integer(dat_train$label) 

if(run.cv.randomForestWeight){
  res_cv_randomForestWeight <- matrix(0, nrow = length(ntree), ncol = 2)
  for (i in 1:length(ntree)){
cat("ntree =", ntree[i],"\n")
res_cv_randomForestWeight[i,] <- cv.randomForestWeight.function(features = feature_train, labels = label_train,K,ntree = ntree[i])
}
  save(res_cv_randomForestWeight, file="../output/res_cv_randomForestWeight.RData")
}else{
  load("../output/res_cv_randomForestWeight.RData")
}

ntree = 50 
ntree = 100 
ntree = 150 
ntree = 200 
ntree = 250 

res_cv_randomForestWeight <- as.data.frame(res_cv_randomForestWeight)
colnames(res_cv_randomForestWeight) <- c("mean_error", "sd_error")
res_cv_randomForestWeight$ntree = as.integer(ntree)
res_cv_randomForestWeight

if(run.cv.randomForestWeight){
plot_meanError_randomForestWeight <- res_cv_randomForestWeight %>%
ggplot(aes(x = as.factor(ntree), y = mean_error,
ymin = mean_error - sd_error, ymax = mean_error + sd_error)) +
geom_crossbar() +
theme(axis.text.x = element_text(angle = 90, hjust = 1))
}
print(plot_meanError_randomForestWeight)

if(run.cv.randomForestWeight){
  ntree_best_randomForestWeight <- res_cv_randomForestWeight$ntree[which.min(res_cv_randomForestWeight$mean_error)]
}
save(ntree_best_randomForestWeight,file = "../output/ntree_best_randomForestWeight.Rdata")
cat("ntree_best_randomForestWeight=", ntree_best_randomForestWeight)

ntree_best_randomForestWeight= 50

###Step 5: Run test on test images

# training weights
weight_train <- rep(NA, length(label_train))
for (v in unique(label_train)){
  weight_train[label_train == v] = 0.5 * length(label_train) / length(label_train[label_train == v])
}
##Training
tm_train=NA
tm_train <- system.time(fit_train_randomForestWeight <- train_randomForest(feature_train, label_train, ntree=ntree_best_randomForestWeight))
save(fit_train_randomForestWeight, file="../output/fit_train_randomForestWeight.RData")

##Testing
tm_test=NA
if(run.test.randomForestWeight){
  load(file="../output/fit_train_randomForestWeight.RData")
  tm_test <- system.time(label_pred_randomForestWeight <- ifelse(test_randomForest(fit_train_randomForestWeight, feature_test)== 2, 1, 0))
}

• evaluation

test_label <- dat_test$label
accurancy.randomForestWeight <- sum(weight_test * (test_label == label_pred_randomForestWeight))/sum(weight_test)
tpr.fpr <- WeightedROC(as.numeric(label_pred_randomForestWeight), test_label)
auc.randomForestWeight <- WeightedAUC(tpr.fpr)

cat("The accuracy of random forest with weight model:", randomForestWeight_model_labels[which.min(res_cv_randomForestWeight$mean_error)], "is", accurancy.randomForestWeight*100, "%.\n")

The accuracy of random forest with weight model: randomForestWithWeight with number of trees = 50 is 54.16667 %.

cat("The AUC of random forest with weight model:", randomForestWeight_model_labels[which.min(res_cv_randomForestWeight$mean_error)], "is", auc.randomForestWeight, ".\n")

The AUC of random forest with weight model: randomForestWithWeight with number of trees = 50 is 0.5416667 .

Summarize Running Time

Prediction performance matters, so does the running times for constructing features and for training the model, especially when the computation resource is limited.

cat("Time for constructing training features=", tm_feature_train[1], "s \n")

Time for constructing training features= 1.237 s 

cat("Time for constructing testing features=", tm_feature_test[1], "s \n")

Time for constructing testing features= 0.353 s 

cat("Time for training random forest with weightmodel=", tm_train[1], "s \n") 

Time for training random forest with weightmodel= 80.535 s 

cat("Time for testing random forest with weight model=", tm_test[1], "s \n")

Time for testing random forest with weight model= 0.118 s